A continuing goal of integrated circuit fabrication is to decrease the amount of semiconductor real estate consumed by integrated circuit devices, and to thereby increase the level of integration.
Memory may utilize a large array of memory devices, with each memory device storing one or more data bits. Accordingly, reduction in the size of individual memory devices may translate into a large increase in the bit density. Common memory devices are dynamic random access memory (DRAM) devices, static random access memory (SRAM) devices, and nonvolatile devices (so-called flash devices). The nonvolatile devices may be incorporated into NAND or NOR memory array architectures.
The size of a memory device may be expressed in terms of the smallest feature size utilized in fabrication of the memory device. Specifically, if the smallest feature size is designated as “F”, the memory device dimensions may be expressed in units of F. Conventional DRAM memory frequently comprises dimensions of at least 6F2, and SRAM may require even more semiconductor real estate.
A type of memory that potentially consumes very little semiconductor real estate is so-called cross-point memory. In cross-point memory, a memory cell occurs at overlap between a wordline and a bitline. Specifically, a memory element material is provided between the wordline and bitline. The memory element material comprises one or more substances which undergo stable and detectable change upon exposure to current; and may be, for example, a perovskite material, a chalcogenide material, an ionic transport material, a resistive switching material, a polymeric material and/or a phase change material. Since the memory cell may be confined to a region of overlap of a bitline and wordline, the memory cell may be theoretically formed to dimensions of 4F2 or less.
It is desired to develop improved methods for forming cross-point memory; and to develop improved cross-point memory structures.